You are on page 1of 27

Summer Project Report

“Molecular Dynamic simulation of a neat Lennard-Jones


fluid and a Lennard-Jones binary mixture”

N. Sridhar
Summer Fellow (2006)
Indian Academy of Sciences

Supervisor: Prof Biman Bagchi


Solid State and Structural Chemistry Unit
Indian Institute of Science
Bangalore
May 8 – July 12

1
Certificate

The work embodied in this report is the result of investigations carried out by Sri N. Sridhar in
Solid State Chemistry Unit, Indian Institute of Science, Bangalore under my supervision
during May 8 to July 15, 2006. Sri Sridhar was a Summer Fellow of the Indian Academy of
Sciences during this period.

Prof Biman Bagchi


Solid State and Structural Chemistry Unit
Indian Institute of Science
Bangalore

2
Acknowledgements

I am grateful to Prof. Biman Bagchi for giving me the opportunity to work in the Solid
State Structural Chemistry Unit of the Indian Institute of Science, Bangalore as a Summer Fellow
of the Indian Academy of Sciences, for introducing me to the exciting area of theoretical chemistry
in general and molecular dynamics in particular, and for overall guidance during the course of this
Fellowship. I am also grateful to Prof S.Yashonath of the Solid State Structural Chemistry Unit for
introducing me to new models and software for simulating molecular dynamics in binary mixtures
and for thought provoking discussions at various stages of the work.

I am grateful to the members of the theoretical and computational chemistry group in the Solid
State and Structural Chemistry Unit for their help and guidance. In particular, I would like to
express my gratitude to Mr.Subrata Pal for his patience in leading me through the initiation phase
and help in subsequent stages in spite of his busy schedule. I also thank Mr.Dwaipayan
Chakrabarti, Ms.Sangeeta Saini, Mr.Bharat Adkar, and Mr.Biman Jana for their cooperation.

I would like to thank my teachers at St. Stephen’s college especially Dr.S.V.Easwaran for
encouraging me to explore science beyond the syllabus.

Finally I express my gratitude to the Indian Academy of Sciences for offering me the Summer
Fellowship and providing me the opportunity to work at an advanced level in the Soild State
Structural Chemistry Unit, Indian Institute of Science, Bangalore.

N.Sridhar

3
Table of Contents

1. Introduction 5

2. Molecular Dynamics
2.1 Conceptual basis 6
2.2 The Molecular dynamics (M.D) simulation 10
2.3 Properties which make MD unique 12
2.4 Binary mixtures 13
2.5 Room Temperature ionic liquids 16
2.6 Limitations of molecular dynamics 17

3. Methodology
3.1 The simulation process and outputs 17
3.2 Systems simulated 23

4. Results
4.1 Neat fluid 26
4.2 Binary mixture 26

5. Conclusion 26

6. References 27

7. Plots 28

4
1. Introduction

Simulation of molecular dynamics allows us to obtain accurate information about the


relationships between the bulk properties of matter and the underlying interactions among the
constituent atoms or molecules in the liquid, solid or gaseous state. Computer simulations provide
a direct route from the microscopic details of the system (the masses of atoms, the interactions
between them, molecular geometry, etc.) to the macroscopic properties of experimental interest
(the equation of state, transport coefficients, structural order parameters, and so on). In addition,
such simulations can be used to simulate critical conditions that are difficult to conduct in real
experiments such as investigations under very high pressure and temperature. Further, the ever-
increasing power of computers also makes it possible to obtain ever more accurate results about
larger and larger systems. As a result, applications of molecular dynamics are to be increasingly
found not only in all branches of chemistry, but also in physics, biophysics, materials science and
engineering, and in the industry as well.

One such area of recent interest is room-temperature ionic liquids (RTILs). RTILs
have attracted tremendous interest in recent years as promising media, which could be an
alternative to the environment polluting, volatile common organic solvents. These are organic
liquids formed solely of ions which, in contrast to their inorganic counterparts like NaCl, exhibit
significantly lower melting temperatures. Besides being liquid at room temperature, they are non-
volatile. This implies that many industrially relevant processes can be influenced significantly if
the dynamics of interfaces between RTILs and other types of liquids are understood. This study
had its motivation in attempting to understand these dynamics through computer simulation.

In view of the limited duration of the Summer Project, and the need to develop a basic
understanding of the molecular simulation processes before proceeding to address the more
complex questions involved in RTILs , the specific objectives of the study were defined as:

(i) develop a basic understanding of simulation of molecular dynamics and related


computer models

5
(ii) apply the principles to molecular dynamics to two simple systems: (a) neat Lennard-
Jones Fluid and (b) a Lennard - Jones binary mixture

The programme code for (a) above was written in FORTRAN 90 and for (b), dl_poly
2 software available in the SSCU was used.

2. Molecular Dynamics

2.1 Conceptual basis:

Molecular dynamics predicts atomic trajectories by direct integration of the equations


of motion – Newton’s second law for classical particles – with appropriate specification of an
inter-atomic potential and suitable initial and boundary conditions.

Treating the problem as the classical many-body problem:

For a system of N particles enclosed in a region of volume V at temperature T, the


positions of the N particles are specified by a set of N vectors, {r(t)} = (r1(t), r2(t),..., rN (t)), with r
j

(t) being the position of particle j at time t. Knowing {r(t)} at various time instants means that the
trajectories of the particles are known.

If the system of particles has a certain energy E which is the sum of kinetic and potential energies
of the particles, E = K + U, where K is the sum of individual kinetic energies:

1 N
K = m∑ vf2
2 j=1

and U is the prescribed interatomic potential given by

U = U (r 1, r2,...,rN ).

6
In general, U depends on the positions of all the particles in a complicated fashion.
We will soon introduce a simplifying approximation (assumption of pair wise interaction) which
makes this most important quantity much easier to handle.

To find the particle trajectories requires the solution of Newton's equations of


motion, F = ma which all particles must satisfy. However, the equations for motion for an N-
particle system are more complicated because the equation for one particle is coupled to all the
other equations through the potential energy U.

The equation that needs to be solved is:

d 2 rj
m = −∇ rj U ( r ) , j = 1,....., N
dt 2
-------(1)

Eq.(1) is a system of second-order, non-linear ordinary differential equations and r1 represents the
famous many-body problem which can be solved only numerically when N is more than 2 to
obtain the atomic trajectories. For this purpose, the time interval is divided into many small
segments, each of Δt. Given the initial condition at time to, {r(to)} , integration means advancing
the system by increments of Δt ,

{r(to)} → {r(to +Δt)} → {r(to + 2Δt)} →...{r( to + Nt Δt)} ------(2)

where Nt is the number of time steps making up the interval of integration.

By Taylor series expansion, for a particle j

1 2
rj (to +Δt) = rj (to ) + vj (t) Δt + aj (t)( Δt) +... ------ (3)
2

Write a similar expansion for rj (to −Δt ) , then add the two expansions to obtain

2
r (to +Δt) =−r (to −Δt) + 2 r (to ) + a ( to )(Δt) +...
j j j j
------ (4)

7
Notice that the left-hand side represents the position of particle j (the trajectory needed) at the next
time step Δ t , whereas all the terms on the right-hand side are quantities evaluated at time t0 and
are therefore known. Eq.(4) is therefore the integration of (1). The acceleration of particle j at time
t0 is F ({r(to )}) / m. The process can be repeated to move another step, etc. and repeated as many
j

times as one wants to generate a sequence of positions (or trajectories) for as long an interval as
desired. There are more elaborate and standard ways of doing this integration as in formal
numerical methods but the basic idea of marching out in discrete steps is the same. A more
accurate method allows one to take a larger value of Δt, which is certainly desirable, but this also
means one needs more memory relative to the simpler method.

These trajectories (positions and velocities) are therefore the raw output of molecular dynamics
simulation. The flow- chart for a typical MD simulation looks like the following.

(a) → (b) → (c) → (d) → (e) → (f) → (g)

a = set particle positions; b = assign particle velocities; c = calculate force on each particle; d =
move particles by time step Δt; e = save current positions and velocities; f = if preset no. of time
steps is reached, stop, otherwise go back to (c); g = analyze data and print results

The Lennard-Jones Pair Potential

To make the simulation tractable, it is common to assume the inter-atomic potential U can be
represented as the sum of pair wise interactions,

1 n
U ( r1 ,....., rN ) ≅ ∑ V ( rij )
2 i≠ j
------ (5)

where rij is the separation distance between particles i and j. V is the pair potential of interaction; it
is a central force potential, being a function only of the separation distance between the two
particles. A very common pair potential used in atomistic simulations is one which describes the
van der Waals interaction in an insulator. This is of the form, Lennard-Jones (L-J or 6-12)
potential) :

8
⎛ ⎛ σ ⎞6 ⎛ σ ⎞12 ⎞
V (r) = 4∈⎜⎜ ⎟ − ⎜ ⎟ ⎟ ------(6)
⎜⎝ r ⎠ ⎝ r ⎠ ⎟
⎝ ⎠
with parameters, ∈ =well-depth σ = hard-sphere diameter

These parameters can be fitted to reproduce experimental data or deduce from results of accurate
12 6
⎛1⎞ ⎛1⎞
quantum chemistry calculations. The ⎜ ⎟ term describes the repulsive force and the ⎜ ⎟ term
⎝r⎠ ⎝r⎠
describes the attractive force.

Thus, by (6), neutral atoms and molecules are subject to two distinct forces in the
limit of large distance, and short distance: an attractive van der Waal's force, or dispersion force, at
long ranges and a short range repulsion force. The repulsion arises from overlap of the electron
clouds, the result of overlapping electron orbitals, referred to as Pauli's repulsion. The attraction is
associated with the interaction between the induced dipole in each atom (the London dispersion
interaction). The Lennard-Jones potential (L-J or 6-12 potential) is a simple mathematical model
that represents this behaviour.

The short-range repulsion rises sharply (with inverse power of 12) at close
interatomic separations, and an attraction varying with the inverse power of 6 (Fig 1). The value of
12 for the first exponent has no special significance, as the repulsive term could just as well be
represented by an exponential, whereas the second exponent results from quantum mechanical
calculations and therefore should not be modified. The importance of short-range repulsion is that
this is necessary to give the system a certain size or volume (density), without which the particles
can collapse onto each other, whereas the attraction is necessary for cohesion of the system of
particles, without which the particles will all fly away from each other. Both are necessary for
solids and liquids to have their known physical properties.

9
Fig.1: Lennard-Jones potential for argon dimer.

Note that the L-J potential is approximate and the form of the repulsion term has no theoretical
justification (the repulsion force depends exponentially on the distance). The attractive long-range
potential, however, is derived from dispersion interactions.

The L-J potential is a relatively good approximation and due to its simplicity often used to describe
the properties of gases, and to model dispersion and overlap interactions in molecular models. It is
particularly accurate for noble gas atoms and is a good approximation at long and short distances
for neutral atoms and molecules.

2.2. The Molecular dynamics (M.D) simulation:

The simulation system is typically a cubical cell in which particles are placed either in
a very regular manner, as in modeling a crystal lattice, or in some random manner, as in modeling
a gas or liquid. The number of particles in the simulation cell is chosen to be quite small. The next
step is to choose the system density. Choosing the density is equivalent to choosing the system
volume since density n = N/V, where N is the number of particles and V is the volume. My code
uses dimensionless reduced units, so reduced density (DR) has typical values around 1.0-1.2 for
solids, and 0.6 - 0.85 for liquids. For reduced temperature (TR) the recommend values are 0.4 - 0.7

10
for solids, and 0.9 - 1.3 for liquids. Assigning particle velocities in (b) above is equivalent to
setting the system temperature.

For simulation of bulk systems (no free surfaces) it is conventional to use the periodic
boundary condition . This means that the cubical cell is surrounded by 26 identical image cells. For
every particle in the simulation cell, there corresponds an image particle in each image cell. The 26
image particles move in exactly the same manner as the actual particle, so that if the actual particle
should happen to move out of the simulation, the image particle in the image cell opposite to the
exit side will move in (and becomes the actual particle, or the particle in the simulation cell) just as
the original particle moves out. The net effect is that with periodic boundary conditions, particles
cannot be lost or gained. In other words, the particle number is conserved, and if the simulation
cell volume is not allowed to change, the system density remains constant.

Since in the pair potential approximation, the particles interact two at a time, a
procedure is needed to decide which pair to consider among the pairs between actual particles
and between actual and image particles. The minimum image convention is a procedure where
one takes the nearest neighbor to an actual particle, regardless of whether this neighbour is an
actual particle or an image particle. Another approximation that is useful to keep the
computations to a manageable level is to introduce a force cutoff distance beyond which particle
pairs simply do not see each other. In order not to have a particle interact with its own image, it is
necessary to ensure that the cutoff distance is less than half of the simulation cell dimension.

Another device often used in MD simulation is a Neighbor List that keeps track of the
nearest, second nearest, ... neighbors for each particle. This is to save time from checking every
particle in the system every time a force calculation is made. The list can be used for several time
steps before updating. Each update is expensive since it involves NxN operations for an N-particle
system. In low-temperature solids where the particles do not move very much, it is possible to do
an entire simulation without or with only a few updating, whereas in simulation of liquids,
updating every 5 or 10 steps are quite common.

11
2.3. Properties which make MD unique:

Classical MD simulation described above (as opposed to quantum MD simulation) is


a useful simulation technique because it follows the atomic motions according to the principles of
classical mechanics as formulated by Newton and Hamilton. Because of this, the results are
physically as meaningful as the potential U that is used. This means that whatever mechanical,
thermodynamic, and statistical mechanical properties that the system of N particles should have,
they are still present in the data. How these properties are extracted from the output of the
simulation – the atomic trajectories – will determine how useful the simulation is. Before any
conclusions can be made, one needs to get in the question of how various properties are to be
obtained from the simulation data. Thus, an MD simulation can be visualized as an ‘atomic video’
of the particle motion (one which we can see as a movie), there is a great deal of realistic details in
the motions themselves, but how to extract the information in a scientifically useful is up to the
viewer. And an experienced viewer can get much more useful information than an inexperienced
one! Besides the above, the following aspects of MD make it very useful and practicable:

(a) Unified study of all physical properties: Using MD one can obtain thermodynamic, structural,
mechanical, dynamic and transport properties of a system of particles which can be a solid,
liquid, or gas. One can even study chemical properties and reactions which are more difficult
and will require using quantum MD.

(b) Several hundred particles are sufficient to simulate bulk matter. While this is not always true, it
is rather surprising that one can get quite accurate thermodynamic properties such as equation
of state in this way. This is an example that the law of large numbers takes over quickly when
one can average over several hundred degrees of freedom.

(c) Direct link between potential model and physical properties. This is really useful from the
standpoint of fundamental understanding of physical matter. It is also very relevant to the
structure-property correlation paradigm in materials science.

(d) Complete control over input, initial and boundary conditions. This is what gives physical
insight into complex system behavior. This is also what makes simulation so useful when
combined with experiment and theory.

12
(e) Detailed atomic trajectories. This is what one can get from MD, or other atomistic simulation
techniques, that experiment often cannot provide. This point alone makes it compelling for the
experimentalist to have access to simulation.

2.4 Binary mixtures:


Binary mixtures are well known to show marked departure from the ideal behavior
given by Raoult’s law. For a given property P, the latter predicts the following simple dependence
on the composition,
P = x1P1 + x2P2 (1)
where xis are the mole fractions and Pis are the values of property P of the pure (single component)
liquids more often than not significant deviations from eqn. (1) is observed which is usually
denoted by an excess function,
Pex = P – (x1P1 + x2P2 ) (2)
Considerable literature exists on such behavior, where P can be volume, free energy
or viscosity. The deviation from ideality appears to have a correlation with the solute-solvent
mutual interaction. Despite the importance and the long interest in this problem, there does not
seem to exist a satisfactory explanation of this non-ideality (or non-additivity) in binary mixtures.
In fact, we are not aware of any microscopic study (based on time correlation function approach)
of the anomalous (or non-monotonous) composition dependence of viscosity. This is, however, not
surprising because a microscopic calculation of viscosity is quite difficult. In the absence of any
microscopic theory, the experimental results have often been fitted to several empirical forms.
Prominent among them is Eyring’s theory of viscosity extended to treat binary mixtures. This
theory can correlate (with the help of one adjustable parameter) several aspects of the composition
dependence of viscosity of many liquid mixtures (like benzene + methanol, toluene + methanol
etc). However, the very basis of Eyring’s theory has been questioned, as this theory is based on
creation of holes of the size of the molecules which is energetically unfavorable.
There also exist several other empirical expressions which attempt to explain the anomalous
dependence of viscosity in binary mixtures. On the experimental side there are evidences of the
correlation between excess viscosity and excess volume of the liquid mixture where it has been
observed for many cases that if excess volume is positive then excess viscosity becomes negative
and vice versa.

13
In recent years interesting theoretical and computer simulation studies on Lennard-Jones (LJ)
binary mixtures have been carried out. These studies have mainly concentrated on the glass
transition in binary mixtures which are known to be good glass formers (in contrast to one of the
component LJ liquid which does not form computer glass easily).In addition, these studies have
considered only one particular composition and a unique interaction strength. Considerable
research has also been carried out by using equilibrium molecular dynamic simulation to determine
the transport properties such as self and mutual diffusion coefficients in binary mixtures.

Non-equilibrium molecular dynamic ~MD simulation methods have also been


employed to determine the shear viscosity and thermal conductivity of binary soft-sphere mixtures.
Heyes carried out the extensive equilibrium MD simulations of Lennard- Jones binary mixtures by
using both the microcanonical ~N V E and canonical ~N V T ensemble methods to study the partial
properties of mixing and transport coefficients by adopting the time correlation function approach.
Apart from the bulk viscosity, these simulations seem to have satisfactorily reproduced in the
experimentally determined transport coefficients for the Ar–Kr mixture.
However, the strong non-ideality in the composition dependence of viscosity,
observed in many experiments, has not been addressed to in the work of Heyes or by others.

14
Fig.2: Composition dependence of excess viscosity and excess volume for a binary mixture with
weak solute-solvent interactions(Model-2).

The non-ideality in the case of inert gas mixtures is small, since their mutual
interaction strength follows the Berthelot mixing rule. To capture this strong non-ideality we
introduce and study two models referred to as model I and model II of binary mixtures in which
the solute–solvent interaction strength is varied by keeping all the other parameters unchanged.

15
All the three interactions solute–solute, solvent–solvent, and solute–solvent are described by the
Lennard-Jones potential. Among the two models, model I promotes the structure formation
between solute and solvent molecules due to strong solute–solvent attractive interaction. The
second model model leads to the opposite scenario by promoting the structure breaking, because of
weak solute–solvent interaction. These two models are perhaps the simplest models to mimic the
structure making and structure breaking in binary mixtures. For convenience, we denote the
solvent molecules as A, and the solute molecules as B.

2.5: Room-temperature ionic liquids:

Room temperature ionic liquids are solvents with many potential applications, as they
have vanishingly low vapour pressures and can be recycled after use in organic reactions. Many
types of chemical reaction can be carried out successfully in these solvents, and solvent recovery
and the work-up of products often depends on their relative solubility in different phases. Thus it is
important to understand the solvation properties of simple solutes in this unusual environment.
Atomistic simulation has proved to be a useful technique for helping to understand chemical
reactivity and solvation properties in aqueous solutions strongly solvated, principally by forming
hydrogen bonds with the chloride ion, while the non-hydrogen bonding solutes interact more
strongly with the cation.

Room temperature ionic liquids (RTILs) have emerged as a cleaner alternative to the
conventional solvents. These are a class of organic salts that, in their pure state, are liquids at or
near room temperature. Some of the more common air-stable ILs are composed of heterocyclic
imidazolium or pyridinium cations having alkyl substituent groups and bulky anions such as
[PF6] − , [BF4] − or [NO3] − .

One of the major barriers preventing the adoption of ILs by industry is the dearth of
physical property data for these compounds, as well as a general lack of fundamental
understanding of how these properties depend on the chemical constitution of the IL. Atomistic
simulation of ILs can help resolve this crisis.

16
2.6. Limitations of molecular dynamics:

There are also significant limitations to MD. The two most important ones are:

(a) Need for sufficiently realistic interatomic potential functions U: This depends on what is
known fundamentally about the chemical binding of the system under study. Progress is being
made in quantum and solid-state chemistry, and condensed-matter physics; these advances
will make MD more and more useful in understanding and predicting the properties and
behavior of physical systems.

(b) Computational capabilities constraints. No computers will ever be big enough and fast
enough. On the other hand, things will keep on improving as far as we can tell. Current limits
on how big and how long are a billion atoms and about a microsecond in brute force
simulation.

3. Methodology

3.1 The simulation process and outputs:

1. Initialisation: To start the simulation we must assign initial positions,


velocities and accelerations to all the particles in the system. I have assumed
the atoms to exist in an f.c.c lattice and accordingly generated the coordinates. I
have also assumed a static system with all atoms at rest initially.
2. Evolving the sample from zero time: The system is evolved over a period
of time (1000 time steps in this case) .First from the L.J potential, we generate
corresponding values for force after each time step. The integration of the
equations of motion is carried out using the Velocity Verlet integration
algorithm.
The Velocity Verlet Algorithm: Positions and velocities at time t = t ++ t are
given by

h2
ri ( t ++t ) = r ( t ) + hvi ( t ) + a (t) --------------------- (1)
2 i

17
+t ⎡
vi ( t ++t ) = vi ( t ) + ⎣
a i ( t ++t ) + a i ( t )⎤⎦ ----------------(2)
2
Fig.2: Algorithm for
an MD simulation

It can be shown that the errors in this algorithm are of (+t ) , and that it is very stable
4

in MD applications and in particular conserves energy very well.

The Verlet algorithm reduces the level of errors introduced into the integration by
calculating the position at the next time step from the positions at the previous and current time
steps, without using the velocity. Velocities, though not required for calculating trajectories are
useful for estimating Kinetic Energy.

This can create technical challenges in molecular dynamics simulations, because


kinetic energy and instantaneous temperatures at time zero cannot be calculated for a system until
the positions are known at time t + ∂ t .

3. Equilibration: A time step ‘t’ is chosen, and the equations of motion are solved
iteratively for a sufficient number of steps to allow the system to come to equilibrium.

4. Simulation: The iterations are continued for a specified period. Physical quantities are measured
at each time step, and their thermal averages are computed as time averages.

This approximation works well for a wide range of materials. Only when we consider
translational or rotational motions of light atoms of molecules (He, Hydrogen, etc) or vibrational
motion with a particular frequency , quantum effects must be considered.
Using the above mentioned algorithm, the velocities and coordinates and of each
particle can be calculated at each time step.

5. Outputs: With the above process, the following physical quantities were derived as functions of
time.

18
Table 1: Equations used for calculating various physical quantities

Physical quantity Equation used

3 m N 2
1. Temperature
Law of Equipartition of energy: N k BT =
2
∑ vi
2 i =1

m N 2
2. Kinetic Energy
K.E = ∑ vi
2 i =1

P.E = ∑ U (| r − r |)
i j
3. Potential Energy ij

Sum of potential and Kinetic energy:


m n 2
( )
4. Total Energy
E = ∑ vi + ∑ U | ri − rj |
2 i=1 ij

N
1
5. Pressure
The Virial theorem: PV = NK BT −
3
∑ r .F
i =1
i i

1 N
( )
6. Velocity Autocorrelation
Cv ( t = n+t ) = * ∑ Vi ( t = t 0 ) .Vi ( t = t 0 + n+t )
Function N i=1

Velocity autocorrelation function( VAF): The velocity autocorrelation function (VAF) is a time
dependent correlation function, and is important because it reveals the underlying nature of the
dynamical processes operating in a molecular system. It is defined as

Cv ( t = n+t ) =
1 N
N ∑
*
i =1
(
Vi ( t = t 0 ) .Vi ( t = t 0 + n+t ) )

Consider a single atom at time zero. At that instant the atom ‘i’ will have a specific
velocity vi. If the atoms in the system did not interact with each other, the Newton's Laws of
motion tell us that the atom would retain this velocity for all time. This means that all the points

19
Cv(t) have the same value, and if all the atoms behaved like this, the plot would be a horizontal
line. It follows that a Velocity autocorrelation plot that is almost horizontal. This implies very
weak forces are acting in the system.

However if the forces are small but not negligible, then the magnitude and direction
of velocity change gradually under the influence of these weak forces. In this case we expect the
scalar product of Vi ( t = t 0 ) with Vi ( t = t 0 + n+t ) to decrease on average, as the velocity is

changed. (i.e. the velocity decorrelates with time, which is the same as saying the atom 'forgets'
what its initial velocity was.) In such a system, the VAF plot is a simple exponential decay,
revealing the presence of weak forces slowly destroying the velocity correlation. Such a result is
typical of the molecules in a gas.

Strong interatomic forces are most evident in high-density systems, such as solids
and liquids, where atoms are packed closely together. In these circumstances, the atoms tend to
seek out locations where there is a near balance between repulsive forces and attractive forces,
since this is where the atoms are most energetically stable. In solids these locations are extremely
stable, and the atoms cannot escape easily from their positions. Their motion is therefore an
oscillation; the atom vibrates backwards and forwards, reversing their velocity at the end of each
oscillation. If we now calculate the VAF, we will obtain a function that oscillates strongly from
positive to negative values and back again. The oscillations will not be of equal magnitude
however, but will decay in time, because there are still perturbative forces acting on the atoms to
disrupt the perfection of their oscillatory motion. So what we see is a function resembling a
damped harmonic motion.

Liquids behave similarly to solids, but now the atoms do not have fixed regular
positions. A diffusive motion is present to destroy rapidly any oscillatory motion. The VAF
therefore may perhaps show one very damped oscillation (a function with only one minimum)
before decaying to zero. In simple terms this may be considered a collision between two atoms
before they rebound from one another and diffuse away.

Uses of VAF: The VAF may be Fourier transformed to project out the underlying frequencies of

20
the molecular processes. This is closely related to the infra- red spectrum of the system, which is
also concerned with vibration on the molecular scale. Also, provided the VAF decays to zero at
long time, the function may be integrated mathematically to calculate the diffusion coefficient D0
given by

α
1
D0 = ∫ vi ( 0) .vi ( t ) dt
30

In addition to the above, the ensemble averages, and the fluctuations were also assessed

Ensemble Averages:

The ensemble is a central concept in statistical mechanics. Imagine that a given


molecular system is replicated many times over, so that we have an enormous number of copies,
each possessing the same physical characteristics of temperature, density, number of atoms and so
on. Since we are interested in the bulk properties of the system, it is not necessary for these
replicas to have exactly the same atomic positions and velocities. In other words the replicas are
allowed to differ microscopically, while retaining the same general properties. Such a collection of
replicated systems is called an ensemble.

Because of the way the ensemble is constructed, if a snapshot of all the replicas is
taken at the same instant, we will find that they differ in the instantaneous values of their bulk
properties. This phenomenon is called fluctuation. Thus the true value of any particular bulk
property must be calculated as an average over all the replicas. This is what is meant by an
ensemble average, and the instantaneous values are said to fluctuate about the mean value.

Molecular dynamics proceeds by a numerical integration of the equations of motion.


Each time step generates a new arrangement of the atoms (called a configuration) and new
instantaneous values for bulk properties such as temperature, pressure, configuration energy etc.
To determine the true or thermodynamic values of these variables requires an ensemble average. In
molecular dynamics this is achieved by performing the average over successive configurations

21
generated by the simulation. In doing this we are making an implicit assumption that an ensemble
average (which relates to many replicas of the system) is the same as an average over time of one
replica (the system we are simulating). This assumption is known as the Ergodic Hypothesis.
Fortunately it seems to be generally true, provided a long enough time is taken in the average.
However it has not yet been rigorously proved mathematically.

Fluctuations

Most of the properties that we calculate for a molecular system are averages. All
averages are obtained by summing over many numbers. Thus in practice we expect the average to
show some dispersion - individual contributions are scattered about the mean value. In statistical
thermodynamics this dispersion about the average value is known as fluctuation and it is both a
subtle and important property of all physical systems.

When calculating an ensemble average (of say, pressure at fixed temperature and
density), we take an instantaneous snapshot of a very large set of replicas of the system concerned
and compute the average from the sum of the individual values taken from each replica. Even
though each replica represents the same system at the same pressure, their individual,
instantaneous values differ slightly, because the molecules that bombard the vessel surfaces to
create the pressure are not in synchronisation between each replica and cannot possibly give rise to
precisely the same surface forces at the same instant. Thus, with pressure, we expect some
fluctuation about the mean value and indeed, similar arguments can be made for all the bulk
properties of the system.
Fluctuations are of fundamental importance in statistical mechanics because they
provide the means by which many physical properties of a molecular system can happen. For
instance, the density of a liquid at equilibrium is a fixed, uniform quantity and we feel justified in
considering the system to be isotropic - the same at all points within its bulk. Yet we know that the
molecules in the system are undergoing diffusion and can easily travel throughout the bulk of the
liquid. It is difficult to imagine how this diffusion can take place if the environment each molecule
is in is completely isotropic. If however we consider the density to be fluctuating minutely from
the mean value at different points in the bulk, we can readily see that such fluctuations would
provide a means by which the diffusion may take place. It is a surprising fact, but most of the

22
physical properties of a bulk system are driven by fluctuations, and indeed can be calculated
directly from them. For this reason it is possible to view fluctuations as even more fundamental
than the average value.

3.2 Systems simulated:

Two kinds of systems were studied, Neat Lennard-Jones Fluid and Binary mixtures

1. Neat Lennard-Jones fluid:

(i) A 256 atom Lennard-Jones system was considered and was evolved over a period of 1000 time
steps. Each time step was equivalent to 0.0032 picoseconds.

The units of mass, length and energy were chosen as m = 1, ∈ = 1, and σ = 1.


All plots are for a temperature of 30K.

(ii). The codes for each program were written in FORTRAN 90. Values for the quantities (in table
1) were calculated after every time step using the equations mentioned. The plot was made using
xmgrace software.

(iii). For plotting velocity autocorrelation, the velocities (of each atom in x, y and z directions)
were read after the system attained equilibrium .i.e. the fluctuations were small. Then these were
read in as arrays and manipulated to get Cv.

2. Lennard-Jones Binary mixture:

A series of molecular dynamic simulations at constant pressure (P), temperature (T),


and total number of particles (N) of binary mixtures were carried out by varying the solute mole
fraction from 0 to 1. Temperature and pressure were kept constant using Berendsen NPT ensemble
with thermostat and barostat relaxation times 1.0 and 0.2 respectively. Our model binary system

23
consists of a total 500 (solute(A)+solvent(B)) particles. We used simple Lennard-Jones (lj)
potential, as pair potential of interaction between any two particles, which is given by the Lennard-
Jones potential function, which sets a cutoff radius rc outside which the potential energy is zero.
The particular form of the potential is given by
U(r) = 4ε [(σ/r)12 – (σ/r)6]

where the cutoff distance rc in this particular case has been taken as equal to 2.5σ. Use of above
potential form takes care of the fact that both potential and force are continuous at the cutoff
distance. A and B denote two different particles. We set the diameter (σ) and molar mass (M) of
the solute as that of Ar, and the solvent as that of Xe, for simplicity, i.e σAA = 3.405 Å, σBB = 4.1 Å
and the molar mass of A = 39.95 gm (that of Argon) and molar mass of B= 131.29 gm (that of
Xenon). The solute-solute interaction strength, εAA = 0.99768 KJ/mol, solvent-solvent interaction
strength, εBB = 1.837394 KJ/mol where A and B represent the solute and solvent particles,
respectively. The variation of total energy and volume of the system with change in composition in
the binary mixture i.e. by changing the mole fractions of particles was studied.
The simulations were carried out in two sets :
(1) The interaction strengths εAB (solute-solvent) were varied in three ways (a) εAB = 1.353931
KJ/mol (Lorentz-Berthelot mixing value), (b) εAB =2.707863 KJ/mol (twice the Lorentz-Berthelot
mixing value), and (c) εAB =0.676966 KJ/mol (half the Lorentz-Berthelot mixing value), keeping
the σAB =3.7525 Å (which is equal to the Lorentz-Berthelot mixing value).

(2) The solute-solvent zero potential diameter σAB was varied in three ways (a) σAB = 3.7525 Å
(Lorentz-Berthelot mixing value), (b) σAB = 4.6 Å, (c) σAB = 3.0 Å, keeping the interaction
strength εAB = 1.353931 KJ/mol (Lorentz-Berthelot mixing value).In set (1), there are three cases,
in (a) εAA< εAB < εBB where the value of solute-solvent interaction is in between the solute-solute
and solvent-solvent interaction strengths, in (b) εAA< εBB < εAB where the solute-solvent interaction
strength is bigger than both solute-solute and solvent-solvent interactions, and in (c) εAB< εAA<
εBB, where the solute-solvent interaction strength is lesser than both solute-solute and solvent-
solvent interaction strengths. In set (2), there are also three cases, in (a) σAA< σAB < σBB where the
solute-solvent zero potential diameter is in between the solvent-solvent and solute-solute zero

24
potential diameter, in (b) σAA< σBB< σAB where the solute-solvent zero potential diameter is
greater than both the solute-solute and solvent-solvent zero potential diameters, and in (c) σAB <
σAA< σBB , where the solute-solvent zero potential diameter is lesser than both solute-solute and
solvent-solvent zero potential diameters.
For both the sets, the pressure P equal to 1 atm and the temperature T as 100 K. After
many trial runs to verify the existing results on total energy and volume of one component liquids,
time step of Δt* = 0.001 ps was found suitable. We have dealt with six different solute
compositions, namely 0.0, 0.2, 0.4, 0.6, 0.8, and 1.0, for each case in each set. For each solute
composition, we have equilibrated the system for 500 ps, i.e the number of equilibration steps was
50,000,and performed the production run for another 50,000 steps, which made the total number of
steps 100,000 3. The plot for average volume of this mixture Vs mole fraction was studied.

(iv) Plots:
The following plots were obtained for the simple system considered:

- The plots of all the physical quantities mentioned in Table 1 were obtained with respect to
time.(figures 1 to 5 )
- The velocity autocorrelation function was plotted for the Lennard Jones fluid considered.
(figure 6 )
- All plots are attached at the end of the report.

25
4. Results

4.1 Neat Lennard-Jones fluid:

- The plots of all the physical quantities mentioned in Table 1 were obtained with respect to
time.
- The velocity autocorrelation function was plotted for the Lennard Jones fluid considered.

4.2 Lennard-Jones Binary mixture:


- A number of trial runs for a binary mixture of Argon and Xenon were carried out using the
parameters mentioned in the methodology. However convincing results were not obtained and
further work is required to refine the simulation. This work is expected to provide very useful
insights into the anomalous composition dependence of viscosity, diffusion, and excess volume of
binary mixtures for which a suitable explanation is not available.

5. Conclusion
The current work is aimed at eventually studying various properties of Room-
Temperature ionic liquids using Molecular Dynamics simulation.
- During the tenure of the fellowship, an exhaustive review was carried out on the basics of
molecular dynamics and a simple Lennard-Jones system was simulated. All codes were written in
FORTRAN 90.
- The variation of various physical quantities including velocity autocorrelation was studied with
respect to time step.
- The use of dl_poly 2 software in molecular dynamics simulation was learnt and applied for
studying non-ideal behaviour of binary mixtures.
- The work on binary liquids is expected to provide very useful insights into the anomalous
composition dependence of viscosity, diffusion, and excess volume of binary mixtures.
- The topic of ‘Room-Temperature ionic liquids’ was reviewed.

26
References:
Textbooks:
1. ‘Computer Simulations of Liquids’ - M.P.Allen and D.J.Tidesley
2. ‘Statistical Mechanics’ - Donald.A.McQuarrie
3. ‘Statistical Mechanics’ - Terell L.Hill
4. ‘Physical Chemical Kinetics’ - RS Berry, SA Rice, J. Ross
5. ‘ Theory of Simple Liquids’ - J.P.Hansen and I.R.McDonald
6. ‘Understanding Molecular Simulation’ - D.Frenkel and B.Smit
7. ‘Computer Programming in FORTRAN 90 and 95’ - V.Rajaraman
8. ‘Physical Chemistry’ – P.W.Atkins

Research papers:
9. Welton.T ; Chem.Rev., 99 (8), 2071 -2084, 1999. 10.1021
10. Jindal.K.Shah et al ; Green Chem., 2002, 4, (2), 112-118
11. Peter Vassilev et al ; J.Chem.Phys, Vol. 115, No. 21, pp. 9815-9820, 2001
12. A.Rahman ; Phys. Rev.136, A405-A411 (1964)
13. A.Rahman and F.Stillinger ; J. Chem. Phys. 55, 3336-3359 (1971).
14. B.Bagchi and Arnab Mukherjee; J.Phys.Chem B 2001, 105, 9581-9585
15. D.M.Heyes; J.Chem.Phys. 1992, 96, 2217-2227

Web links:
16. http://www.ccl.net/cca/software/SOURCES/FORTRAN/allen-tildesley -
book/README.shtml
17. http://www.mse.ncsu.edu/CompMatSci/Tutorial/index.html
18. http://lem.ch.unito.it/didattica/infochimica/Liquidi%20Ionici/Definition.html#note
19. http://www.fisica.uniud.it/%7Eercolessi/md/md/node1.html
20. http://www.ch.embnet.org/MD_tutorial/pages/MD.Part1.html

27

You might also like